Michael Burry, the investor made famous by his housing-market bet before 2008 earning him The Big Short nickname, has rebuilt his AI-skeptic portfolio through put options. He has closed most of his direct short positions in AI-linked companies and replaced them with put contracts, which pay off if a stock falls. The change signals that he now expects the correction to arrive earlier than his own earlier forecast.
The new positions are concentrated in the names he has targeted for months:
- Micron: puts expiring in June, with a strike near $500.
- Nebius: puts expiring in June, with a strike in the double digits.
- SOXX (semiconductor ETF): puts expiring September 2027, with a strike in the low $400s.
- Palantir: an enlarged put position expiring September 2027, with a strike around low $100s.
- MetLife: a new bet, using multi-year puts struck in the $70s.
He also exited CoreWeave but says he hasn’t found puts on it at a price he likes. Burry says the premiums are low right now because market volatility gauges are calm, so protection against a fall is unusually cheap. He also acknowledged that some of the rebalancing was about taxes.
The larger driver, by his account, was the research that convinced him the timeline should be shortened. Options only work if the move happens before expiry, so a shorter timeline is what makes this structure attractive. In August his central case put the unwinding in 2028.
The research he leaned on comes from Ares Management. Its point is that much of the AI buildout depends on revenue that hasn’t yet materialized, funded through arrangements with strict contractual terms. If earnings fall short of the spending plans for even a single season, a few decision-makers could redirect capital elsewhere, and the paperwork was drafted with that possibility in mind. On this view, AI doesn’t need to fail. It only needs to underdeliver briefly for money to start moving out.
Burry’s reputation comes from betting against housing before the 2008 crisis, a trade that was correct long before it was profitable. In late 2025, disclosures showed he held bearish option positions on Palantir and Nvidia, and AI stocks were falling around the same time. Some commentators tied the two together. That link is worth treating carefully: rate worries, valuation concerns and profit-taking were all in play, and no one can say how much one investor’s filing contributed.
He has been vocal since then. In May he likened the market’s mood to the last stretch of the dot-com boom, and earlier this month he added to shorts on Micron, Nebius and SOXX while the Nasdaq 100 hit a record. Markets have largely gone against him so far.
The June expiries are the first real test. If Micron and Nebius are not meaningfully lower by then, those contracts lose their value entirely, so the cost of being early is now explicit. The SOXX and Palantir puts buy extra time.
